Wildlife Society Bulletin 45(4):618 – 629; 2021; DOI: 10.1002/wsb.1232 Research Article Plant Community Response and Implications for Wildlife Following Control of a Nonnative Perennial Grass CRAIG A. HARPER , 1 Department of Forestry, Wildlife and Fisheries, University of Tennessee, 2431 Joe Johnson Drive, Knoxville, TN 37996, USA J. WADE GEFELLERS, Department of Forestry, Wildlife and Fisheries, University of Tennessee, 2431 Joe Johnson Drive, Knoxville, TN 37996, USA DAVID A. BUEHLER, Department of Forestry, Wildlife and Fisheries, University of Tennessee, 2431 Joe Johnson Drive, Knoxville, TN 37996, USA CHRISTOPHER E. MOORMAN, Fisheries, Wildlife, and Conservation Biology Program, Department of Forestry and Environmental Resources, North Carolina State University, Raleigh, NC 27695, USA JOHN M. ZOBEL, 2 Department of Forestry, Wildlife and Fisheries, University of Tennessee, 2431 Joe Johnson Drive, Knoxville, TN 37996, USA ABSTRACT Restoration of early successional plant communities dominated by nonnative plant species is a central focus of many state and federal agencies to improve habitat for wildlife associated with these communities. Restoration e ff orts largely have concentrated on controlling nonnative species followed by planting native grasses and forbs. However, there are numerous establishment problems associated with planting that warrant evaluation of alternative approaches for restoration. We conducted a fi eld experiment to compare vegetation composition and structure as related to habitat for focal wildlife among plant communities established by planting (Planted) native grasses and forbs and revegetation from the seedbank (Seedbank) without planting following control of tall fescue ( Schedonorus arundinaceus ) at 15 replicated sites in Tennessee and Alabama, USA. Planted and Seedbank treatments produced similar plant communities. Vegetation structure providing cover for nesting and brooding northern bobwhite ( Colinus virginianus ) and wild turkey ( Meleagris gallopavo ) was similar between Seedbank and Planted treatments except native grass cover was greatest in Planted, and we recorded greater openness at ground level in Seedbank than Planted or tall fescue control (Control). Abundance of northern bobwhite food plants and selected white ‐ tailed deer ( Odocoileus virginianus ) forage were similar between Planted and Seedbank treatments, but nutritional carrying capacity for deer was greatest in Seedbank. Despite similarities in food abundance, and even though all forbs included in the planting mixtures were food plants, the majority of food plants in Planted were from the seedbank. The compositional and structural characteristics deemed most in fl uential in previous studies to selection of breeding sites by dickcissel ( Spiza americana ), fi eld sparrow ( Spizella pusilla ), grasshopper sparrow ( Ammodramus savannarum ), Henslow ’ s sparrow ( Ammodramus henslowii ), and northern bobwhite were similar in Planted and Seedbank. Tall fescue Control was most similar to char - acteristics of eastern meadowlark ( Sturnella magna ) breeding sites. Revegetation following Seedbank pro - duced a plant community that provided habitat for many wildlife species equal to or better than Planted and was 3.7 times less expensive than Planted. © 2021 The Wildlife Society. KEY WORDS conservation programs, early successional communities, grassland restoration, grassland songbirds, northern bobwhite, seedbank, tall fescue, white ‐ tailed deer, wild turkey. Restoration of native grasslands and other early successional communities dominated by forbs, grasses, and other herba - ceous plants representative of an early seral stage is a con - servation focus in the eastern U.S. (Noss et al. 1995, Harper 2017, Keyser et al. 2019). Substantial land ‐ use changes through urbanization, intensi fi ed agriculture, and commercial forestry have reduced native, early ‐ successional plant com - munities (Ramankutty and Foley 1999, Drummond and Loveland 2010), and associated wildlife have experienced steep declines in recent decades (Brennan 1991, Peterjohn and Sauer 1997, Kirkland and Hart 1999, Pruitt 2000, Mcchesney and Anderson 2015). Conservation e ff orts have concentrated on converting row ‐ crop agriculture and non - native grassland to native plant communities, particularly to increase and enhance habitat for conservation ‐ priority species, Received: 15 June 2020; Accepted: 16 May 2021 Published: 8 December 2021 1 E ‐ mail: charper@utk.edu 2 Current a ffi liation: Department of Forest Resources, University of Minnesota, 1530 Cleveland Avenue N, St. Paul, MN 55108, USA 618 Wildlife Society Bulletin • 45(4) such as northern bobwhite ( Colinus virginianus ; hereafter bobwhite) and grasshopper sparrow ( Ammodramus sav - annarum ) (Johnson and Schwartz 1993, Delisle and Savidge 1997, Herkert 1998, Fletcher et al. 2006, Crosby et al. 2015). Conversion of nonnative grasslands to native plant com - munities has been a major objective because of the consid - erable negative e ff ects on the diversity and function of native early successional plant communities (Rudgers and Clay 2007, Barnes et al. 2013). In the eastern U.S., tall fescue ( Schedonorus arundinaceus ) is the most commonly occurring nonnative grass covering approximately 15 million hectares (Ball et al. 2003). Tall fescue is an introduced, sod ‐ forming, cool ‐ season grass widely promoted by forage agronomists and soil conservationists from the 1940s – 1970s, and more recently through the United States Department of Agriculture ‐ Natural Resources Conservation Service ’ s (USDA ‐ NRCS) Conservation Reserve Program (CRP) as a livestock forage and to aid in protection against soil erosion (Carmichael 1997, Rogers and Locke 2013). Tall fescue presents problems for various wildlife because of its growth habit and its association with an endophytic fungus ( Neo - typhodium coenophialum ) that infects > 90% of the grass (Bacon and Siegel 1988, Stuedemann and Hoveland 1988, Ball et al. 2003). The dense structure of tall fescue at ground level suppresses the seedbank, lowering plant diversity (GeFellers et al. 2020); restricts movement of small wildlife species, and lacks vertical structure important for bobwhite chicks, wild turkey ( Meleagris gallopavo ) poults, and nu - merous other ground ‐ feeding wildlife species (Barnes et al. 1995, Washburn et al. 2000, Harper et al. 2007, Barnes et al. 2013). The endophyte fungus compounds the problem by giving tall fescue a competitive advantage over native plant species (Clay 1990, Latch 1993, Hill et al. 1996, Salminen et al. 2005, Rudgers et al. 2010), especially during drought conditions; this grass also releases toxic ergot alkaloids that have negative e ff ects on wildlife (Betsill et al. 1979, Madej and Clay 1991, Clay et al. 1993, Coley et al. 1995, Conover and Messmer 1996). State wildlife agencies across most of the eastern U.S., along with the USDA ‐ NRCS and USDA ‐ Farm Service Agency, actively promote conversion of nonnative grass - lands to native plant communities through conservation programs, such as CRP, Conservation Reserve Enhance - ment Program (CREP), and Environmental Quality In - centives Program (EQIP). The conversion from nonnative to native plant communities is intended to increase and enhance habitat for species of concern that require early successional plant communities (Allen and Vandever 2012, Kentucky Department of Fish and Wildlife Resources 2013, Georgia Department of Natural Resources 2015, Tennessee Wildlife Resources Agency 2015, USDA ‐ NRCS 2018). Although some grassland obligate songbirds, such as eastern meadowlark ( Sturnella magna ), may use fi elds dominated by nonnative grasses for nesting (McCoy et al. 2001, Moorman et al. 2017), conversion to native plant communities o ff ers bene fi ts to a broad suite of species. Species bene fi ting from conversion include conservation ‐ priority species such as grasshopper sparrow and Henslow ’ s sparrow ( Ammodramus henslowii ) as well as species commonly selected for man - agement by private landowners, such as white ‐ tailed deer ( Odocoileus virginianus ) and wild turkey (Nagy ‐ Reis et al. 2019, Reiley et al. 2019, Lituma and Buehler 2020). Enrollment in most state and federal conservation pro - grams requires landowners to plant a native seed mixture following control of nonnative grasses such as tall fescue. Several problems are commonly associated with planting, including improper site preparation and equipment setup, lack of weed control options that will not harm planted species, and high seed costs, which are paid with taxpayer and sportsman dollars through conservation programs (Harper et al. 2007, Kettenring and Adams 2011). An al - ternative approach to circumvent establishment problems and cost may be to relax the planting requirement. Allowing the naturally occurring seedbank to revegetate the site fol - lowing eradication of nonnative species could allow more options for controlling undesirable vegetation (e.g., non - native invasive species), eliminate planting and site prepa - ration issues, and reduce costs because seed purchase would not be necessary. We conducted a fi eld experiment to evaluate plant com - munity response and associated e ff ects on various habitat components for selected wildlife species following tall fescue control and revegetation via planting a native grass/forb mixture and via revegetation from the seedbank without planting. We evaluated e ff ects on vegetation composition and structure for a number of species: 1) 5 grassland song - birds undergoing population decline and identi fi ed as pri - ority species in conservation programs; 2) northern bob - white, which also is undergoing population decline and requires early successional plant communities; and 3) white ‐ tailed deer and wild turkey because they are primary focal species of private landowners in the eastern U.S. Deer and turkeys are considered generalist species, but their habitat may be improved by restoring early successional plant communities. We measured cover of food plants for bob - white and measured forage availability for deer. We hy - pothesized plant community composition following planting would favor native warm ‐ season grasses, whereas revegetation from the seedbank without planting would produce more forb cover, which would increase forage availability for white ‐ tailed deer and food plants for northern bobwhite. We hypothesized that vegetation structure following planting and revegetation from the seedbank would be similar and would be favorable to wildlife that prefer taller structure, but that tall fescue control would provide structure more favorable to songbirds that prefer shorter structure. STUDY AREA We conducted our study across 15 fi elds dominated ( > 75% cover) by tall fescue in Tennessee and northern Alabama, USA (Fig. 1). The fi elds we used historically were hayed or grazed, but each had been idle from haying or grazing for > 10 years prior to our study and maintained by annual mowing, representing standard management of such fi elds, which are ubiquitous in the region (Dykes 2005, Rogers and Harper et al. • Establishing Early Successional Plant Communities 619 23285540, 2021, 4, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1232 by University Of Florida, Wiley Online Library on [06/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Locke 2013). Although the fi elds were dominated by tall fescue, each was undergoing succession with various forbs (e.g., Canada goldenrod [ Solidago canadensis ], wingstem [ Verbesina alternifolia ]), and brambles ( Rubus spp.) pio - neering from the seedbank, providing a di ff erent structure than that present in tall fescue fi elds maintained for hay or pasture production. Fields ranged in size from 2.2 to 5.3 ha (3.4 ± 0.2 [SE], n = 15). Seven study sites were on Tennessee Wildlife Resources Agency property in Cocke, Cumberland, Lawrence, Roane, Union, White, and Williamson counties. Six study sites were located on Tennessee Valley Authority properties in Bedford, Ham - blen, Je ff erson, Monroe, and Sevier Counties, Tennessee, and Franklin County, Alabama. One study site was on Alabama Department of Conservation and Natural Re - sources property in Jackson County, and one was in Cades Cove within the Great Smoky Mountains National Park (hereafter Park) in Blount County, Tennessee. Elevations ranged from 181 m to 658 m. Mean daily temperature across the study area ranged from − 4°C to 33°C, with mean annual precipitation that ranged 114 cm to 152 cm (Na - tional Oceanic and Atmospheric Administration 2019). Soils were classi fi ed as silt loam or silty clay at all sites (USDA ‐ NRCS 2019). METHODS Study Design Our experimental design was a randomized complete block design with replication. We divided each fi eld into 3 similar ‐ sized treatment units, and we randomly assigned 1 of 3 treatments (control [Control], seedbank response without planting [Seedbank], and planted [Planted]) to each unit. No treatment was made to change plant com - munity composition in Control units, and we mowed Control units annually in late winter to represent pretreat - ment conditions throughout the study and represent default management practices common in idle tall fescue fi elds (Dykes 2005). Treatment units varied in size from 0.8 to 1.6 ha (1.1 ± 0.1, n = 45). Treatment initiation. — We mowed all study sites in October 2015 and allowed them to regrow to 15.2 – 25.4 cm (Harper 2017). We then broadcast ‐ sprayed glyphosate (2.8 kg active ingredient (ai)/ha) herbicide in Planted and Seedbank units to control tall fescue in November – December 2015. We used spot ‐ spray glyphosate applications in February – March 2016 to control any tall fescue missed during initial applications. We made herbicide applications when temperatures were at or above 10°C to ensure e ff ectiveness of the herbicide because tall fescue actively grows at temperatures as low as 3°C (Gastal et al. 1992, Rogers and Locke 2013). Planting treatment. — We planted a native warm ‐ season grass and forb seed mixture in Planted treatment units in April – May 2016 (Table S1, available online in Supporting Information). Planted units simulated plantings made on lands enrolled in conservation programs (e.g., CRP and EQIP), and the seed mixture and planting rate were determined by Private Lands Wildlife Biologists with Tennessee Wildlife Resources Agency and Alabama Department of Conservation and Natural Resources who implement their conservation programs. We planted all sites with the same seed mixture excluding the Park site because the National Park Service prohibits introduction of outside genotypic seed sources. Seeds planted at the Park site were Figure 1. Map of 15 study site locations in Tennessee and Alabama, USA, 2016 – 2018, where we compared vegetation composition and structure as related to several components of habitat for focal wildlife. 620 Wildlife Society Bulletin • 45(4) 23285540, 2021, 4, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1232 by University Of Florida, Wiley Online Library on [06/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License collected from within Cades Cove by National Park Service personnel (Table S2, available online in Supporting Information). We used no ‐ till drills (Truax TM Flex II Series drills, Truax Company Inc., New Hope, MN, USA, and Haybuster® drills, Duratech Industries International Inc., Jamestown, ND, USA) to plant seed. We calibrated and adjusted drills to ensure seed were planted at the recommended seeding rate of 7.3 kg/ha pure live seed and that planting depth was ≤ 0.64 cm. We made preemergence imazapic (Plateau®, BASF, Research Triangle Park, NC, USA) applications (0.07 – 0.105 kg ai/ha) within 7 days of planting to control competition (Washburn et al. 1999, Harper et al. 2007). Natural revegetation treatment. — We allowed the seedbank to revegetate Seedbank units following treatment initiation in October 2015. We later used herbicide applications described below as necessary to remove undesirable vegetation and promote a desirable early ‐ successional, native plant community. Herbicide applications in Seedbank and Planted treatment units. — To control undesirable plant species during each year of establishment, we made spot ‐ spray applications using 15 ‐ L backpack sprayers (Solo USA, Newport News, VA, USA) and/or a 95 ‐ L ATV sprayer (Cabelas, Sydney, NE, USA) equipped with a spray gun (Green Garde®, H.D. Hudson Manufacturing Company, Chicago, IL, USA). We used spot ‐ spray applications most often (69% and 86% of all applications made in Seedbank and Planted, respectively). We de fi ned spot ‐ spraying as any herbicide application that did not impact the entire treatment unit. Spot ‐ spray applications on average impacted < 20% of any single treatment unit. Broadcast applications impacted 100% of any single treatment unit (31% and 14% of all applications made in Seedbank and Planted, respectively). We made broadcast applications with a tractor and 3 ‐ point boom sprayers, ATV sprayer with boom attachment, or 4 ‐ nozzle handheld booms (R&D Sprayers, Opelousas, LA, USA). We used broadcast applications during fall/winter when ≥ 50% of the treatment unit was comprised of undesirable cool ‐ season species and during summer when ≥ 90% of a treatment unit was comprised of undesirable warm ‐ season species. We used spot ‐ spray applications otherwise. We determined herbicides, application rates, and application timing based on plant species targeted for removal. All herbicide applications were made in accordance with label recommendations and federal laws governing their use. We recorded the number of herbicide applications made and how much of each herbicide was applied to later calculate average costs. We maintained Planted units throughout the study con - sistent with what is required to remain in compliance with conservation program rules and according to Private Lands Biologists ’ recommendations. If we detected > 30% cover of johnsongrass ( Sorghum halepense ), crabgrass ( Digitaria spp.), or Japanese stiltgrass ( Microstegium vimineum ), we spot ‐ sprayed these species with imazapic because they are con - trolled by this herbicide whereas the planted species are resistant to it. Bermudagrass ( Cynodon dactylon ) also was a common problem in Planted units. We sprayed bermuda - grass regardless of percent cover, and we sprayed woody species (i.e., trees and shrubs) in Planted treatment units if they reached 5% cover. We spot ‐ sprayed undesirable species in Seedbank units on average once/year regardless of coverage. Undesirable veg - etation most often included species identi fi ed by the Southeast Exotic Pest Plant Council as nonnative invasive species. Commonly occurring undesirable nonnative species included johnsongrass, bermudagrass, crabgrass, sericea lespedeza ( Lespedeza cuneata ), narrowleaf plantain ( Plantago lanceolata ), musk thistle ( Carduus nutans ), and common chickweed ( Stellaria media ). The areas opened by herbicide applications revegetated naturally again. Certain native species, including Rubus spp., broomsedge bluestem ( Andropogon virginicus ), and black locust ( Robinia pseudoa - cacia ), also can dominate open areas. To promote plant species diversity, we reduced these species with the appro - priate herbicide application if they exceeded 30% cover, as assessed by our transect data. Response Variables Vegetation composition. — We collected all vegetation data from June – August 2016 – 2018 along 5, 50 ‐ m systematically ‐ spaced transects across and throughout each treatment unit at each site and maintained a minimum 10 ‐ m bu ff er from unit edges. We conducted line ‐ point intercept sampling to quantify vegetation composition in all treatments (Herrick et al. 2009). Every plant species that intercepted each line point along each transect was recorded at 2 ‐ m intervals. We calculated percent cover of species and vegetation life form (bramble, forb, grass, and woody) by dividing the number of hits of each species or life form by the total number of sampling points per transect. We then averaged percent cover of each species or life form across all transects for each treatment to calculate percent cover. Vegetation structure. — We measured visual obstruction of vegetation using a modi fi ed vegetation pro fi le board (Nudds 1977). The vegetation pro fi le board was 2 ‐ m tall and divided into 5 alternating ‐ colored rectangular sections. The bottom 0.5 m was divided into 2, 0.25 ‐ m × 0.25 ‐ m sections, whereas the upper 1.5 m was divided into 3, 0.5 ‐ m × 0.25 ‐ m sections. The bottom 0.25 m represented visual obstruction at the level where bobwhite and other small ground ‐ dwelling wildlife species occur. Visual obstruction 0 – 0.5 m represented that occurring at the upper end of vegetation height important to brooding wild turkey (Metzler and Speake 1985, Peoples et al. 1996, Wood et al. 2019). Huegel et al. (1986) indicated that visual obstruction ≥ 1 m was important at deer fawn bedsites because taller vegetation may maintain more stable temperatures than shorter vegetation. We recorded 2 visual ‐ obstruction meas - urements along each transect 10 m on either side of center. One person knelt at plot center and estimated visual obst - ruction by placing each of the 5 sections into 1 of 6 categories (0 = no vegetation, 1 = 1 – 20%, 2 = 21 – 40%, 3 = 41 – 60%, 4 = 61 – 80%, and 5 = 81 – 100% visual obstruction). We qualitatively compared visual obstruction Harper et al. • Establishing Early Successional Plant Communities 621 23285540, 2021, 4, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1232 by University Of Florida, Wiley Online Library on [06/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License measurements across treatment units to that reported as used by wild turkey, bobwhite, and deer in other studies. We used measurements from a ground ‐ sighting tube to provide an index of openness at ground level (Gruchy and Harper 2014). We recorded 2 sight ‐ tube measurements along each transect at 14 and 34 meters. The observer looked through the sighting tube while another team member placed a 5.1cm ‐ diameter PVC pole in front of the tube. The pole was moved away until the bottom 15 cm was completely obscured by vegetation and then the dis - tance from the pole to the sighting tube was recorded. Openness at ground level is an important habitat com - ponent for several wildlife species, as it allows mobility to access food resources and escape predation (Rosene 1969, Harper et al. 2007). We measured litter depth at the 15 m and 35 m intercept along each transect. We collected measurements using a 30.5 ‐ cm metal ruler to the nearest 0.5 centimeter from mineral soil to the top of accumulated plant litter. Structural characteristics of grassland and shrubland bird breeding sites. — Treatment and control units were too small to monitor actual bird use in response to the establishment practices. However, to evaluate the suitability of the vegetation characteristics in treatment and control units to areas actually selected by birds for breeding, we used the same vegetation composition and structural data from 2 published studies located in the same ecoregion. These studies were conducted in relative proximity to our study sites and identi fi ed variables characteristic of actual breeding sites selected by dickcissel ( Spiza americana ), eastern meadowlark, fi eld sparrow ( Spizella pusilla ), grasshopper sparrow, Henslow ’ s sparrow, and bobwhite. Vegetation characteristics surrounding songbird and bobwhite nest sites were collected at Fort Campbell Army Installation in Hopkinsville, Kentucky, USA (Giocomo 2005, Giocomo et al. 2008) and Peabody Wildlife Management Area in west ‐ central Kentucky (Brooke et al. 2017). Given the area and landscape requirements of the species we considered, we did not use breeding ‐ site data to speculate on bird use of our study fi elds, per se, but rather to aid in comparison of treatment and control units as potential breeding sites by the grassland songbirds and bobwhite. Food availability for bobwhite and forage selectivity by deer. — We classi fi ed bobwhite food plants as those producing seed and/or soft mast commonly consumed by bobwhite (Rosene and Freeman 1988, Johnson et al. 2018). We inspected each plant recorded along point ‐ intercept transects to determine if the plant had been eaten by deer. We divided the number of stems that had been eaten by the number of stems available on that plant to measure deer browse intensity and selectivity using the Chesson index (Chesson 1983, Shaw et al. 2010). A fi fteenth percentile cut ‐ o ff value was used to rank species selection because that cut ‐ o ff value closely matched fi eld observations of deer selectivity and has been used by previous researchers (Nanney et al. 2018). Species determined to be selected by deer were included in nutritional carrying capacity calculations. Deer forage availability. — We randomly placed 2, 1 ‐ m 2 frames along each transect and all vegetation ≤ 2 m in height within each frame was cut with a hedge trimmer (Stihl HS 45, Virginia Beach, VA, USA) at ground level (Lashley et al. 2014). We placed all cut vegetation in a cloth sack and assigned unique labels to each sample. Forage samples were separated into selected and non ‐ selected deer forages. We separated selected forages by species and by young and old plant portions because deer are concentrate selectors and select the youngest and most nutritious portions of plants (Hewitt 2011, Lashley et al. 2014). We dried all forage samples for 72 hours at 50°C in a forced ‐ air oven dryer. We weighed each sample with calibrated digital scales to the nearest 0.1 g. We then packaged samples and shipped them to the Agriculture Service Laboratory at Clemson University for wet chemistry nutritional analysis. Nutritional carrying capacity. — We calculated estimates of nutritional carrying capacity for deer using a mixed ‐ diet approach with nutritional constraints according to Hobbs and Swift (1985). We used a nutritional constraint of 14% crude protein with a 2.4 kg/day intake rate to represent nutritional needs of a 50 ‐ kg doe at peak lactation with twin fawns (National Research Council 2007, Hewitt 2011, Lashley et al. 2011, Nanney et al. 2018). Data Analysis We analyzed data collected in 2018 (third growing season) to compare treatment e ff ects and the resulting habitat quality of established vegetation for the wildlife species considered in our analysis. We used 2018 data because re - storation of native plant communities commonly requires 2 – 3 years (Fransen et al. 2006, Harper et al. 2007, Rushing 2014). We used mowing and broadcast herbicide applica - tions as part of the establishment process during the growing seasons of 2016 – 17, which highly altered vegeta - tion composition and structure during those growing sea - sons, to promote native plant communities in both Seed - bank and Planted treatment units. We fi t analysis of variance (ANOVA) models with blocking using program R (v. 3.5.1, R. Core Team 2018) to detect di ff erences among treatments in percent cover of brambles, forbs, grasses, woody plants, and quail food plants, litter depth, ground ‐ sighting distance, visual obstruction, deer forage availability, and deer nutritional carrying capacity across 15 replicate fi elds at signi fi cance level α = 0.05. We used post ‐ hoc Tukey HSD tests to compare treatment estimates when a signi fi cant e ff ect of treatment was observed. We met as - sumptions of normality and equal variance using arcsine square root transformations on percent cover of brambles, forbs, woody species, and visual obstruction data. Addi - tionally, we used square root transformations on forage availability. We determined vegetation characteristics that best ex - plained selection of breeding sites by grassland and shrub - land birds using multivariate factor analyses (FA) in R. We fi rst performed principal component analysis to de - termine how many factors to include in the FA. We then plotted vegetation characteristics assigned to factors 1 and 2 622 Wildlife Society Bulletin • 45(4) 23285540, 2021, 4, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1232 by University Of Florida, Wiley Online Library on [06/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License by FA on a biplot with 95% con fi dence ellipses around the multivariate centroid for each species. We standardized treatment data from our study with data from the Ft. Campbell and Peabody WMA bird nest datasets and con - ducted identical FA procedures, so results were comparable across the 2 datasets. We plotted factor scores for treat - ments with 95% con fi dence ellipses on the biplot with the bird nest ‐ site factor scores. Using ArcMap 10.5 (ESRI, Redlands, CA, USA), we calculated the percent of each treatment ellipse that was contained within the 95% nest ‐ site ellipses for each species. We plotted factor 1 on the x ‐ axis and factor 2 on the y ‐ axis of biplots for all species, except grasshopper sparrow, for which factor 2 was ex - plained by vegetation height variables that were not meas - ured in our study. We plotted factor 3 instead because it was explained by variables that were collected in our study and explained nearly as much variability (12.8%) as did factor 2 (14.8%). RESULTS Vegetation Composition Tall fescue cover was greatest ( F 2,28 = 213.11, P ≤ 0.001) in Control (75% ± 2.1%), with minimal cover in Seedbank (6% ± 1.1%) and Planted (2% ± 0.6%). Forb cover in Seed - bank (72% ± 2%) and Planted (64% ± 3%) was increased 50% and 33% over that in Control (48% ± 3%), respectively (Table 1). Overall grass cover was greatest in Control (92%). Native warm ‐ season grass cover in Planted (61% ± 2%) and Seedbank (49% ± 4%) was increased 85% and 48% over that in Control (33% ± 4%), respectively. Percent cover of bram - bles and woody species was ≤ 12% and ≤ 9% across all treat - ments, respectively (Table 1). Percent cover of plant species producing bobwhite foods was approximately 30% greater ( F 2,28 = 2.93, P = 0.070) in Seedbank (43% ± 3%) and Planted (41% ± 3%) than Control (32% ± 3%). Vegetation Structure Visual obstruction 0 – 0.25 m above ground did not vary among treatments (all 99 – 100% cover), and it did not vary between Seedbank and Planted from 0.25 m up to 2 m. However, there was less visual obstruction in Control from 0.25 m to 1.5 m than Seedbank and Planted, and less in Control than Planted from 1.5 m to 2.0 m (Table 2). Of particular importance was a minimum 28 – 51% increase in visual obstruction from 0.5 m to 1.5 m in Seedbank and Planted over Control. We detected a treatment e ff ect ( F 2,28 = 4.79, P = 0.016) for ground ‐ sighting distance. Average ground ‐ sighting distance was similar in Planted (66 ± 3 cm) and Control (63 ± 2 cm; P = 0.916), but ap - proximately 30% farther in Seedbank (85 ± 5 cm). Litter depth (Control = 3.5 ± 1.2 cm; Seedbank = 2.6 ± 1.5 cm; Planted = 3.5 ± 2.2 cm) was similar among treatments ( F 2,28 = 2.83, P = 0.076). Suitability of Vegetation Among Treatments for Breeding Birds Ellipses identifying compositional and structural variables most in fl uential to selection by breeding birds were similar between Seedbank and Planted but varied by bird species (Table S3 and Figs. S1 – S6, available online in Supporting Information). Treatment and control ellipses were 94 – 100% contained within the dickcissel and grasshopper ellipses. The Seedbank ellipse was most similar (90%) to the fi eld sparrow ellipse. The Seedbank ellipse also was most similar (82%) to the Henslow ’ s sparrow ellipse. The Control ellipse was most similar (98%) to the eastern meadowlark ellipse with regard to vegetation composition and structure. Seedbank (86%), Control (85%), and Planted (79%) ellipses all were relatively similar to the bobwhite ellipse with regard to vegetation composition and structure. Deer Forage Availability and Nutritional Carrying Capacity We documented 290 plant species across all sites and years. We classi fi ed 14 documented species as moderately and highly selected by deer using a selection index and a cut ‐ o ff value of α = 0.005 (Chesson 1978). Selected species in - cluded 9 forbs, 2 brambles, 2 trees, and 1 vine. No grasses were selected (Table 3). All 14 plant species were included in nutritional carrying capacity calculations. Forage avail - ability did not di ff er ( F 2,28 = 2.49, P = 0.101) among treatments (Seedbank = 570 ± 54 kg/ha; Planted = 452 ± 58 kg/ha; Control = 429 ± 60 kg/ha). The 5 forb species in mixtures seeded in Planted all were considered selected deer forages, but contributed only 26 ± 9 kg/ha, indicating 94.2% of the deer forages in Planted occurred naturally from the seedbank. Nutritional carrying capacity was 2.2 times greater in Seedbank (145 ± 14 deer days/ha) than Control (66 ± 10 deer days/ha, n = 15; P = 0.013) and 1.7 times greater than Planted (88 ± 11 deer days/ha; P = 0.090). Treatment Costs and E ff ort Considering cost for seed and herbicide, the average cost for Planted treatments was $ 468.98 per hectare. Glyphosate applications to prepare Planted treatments were $ 20.26 per hectare, the preemergence imazapic application was $ 16.61 per hectare, seed cost $ 400.38 per hectare, and post ‐ planting herbicides for weed control averaged $ 31.73 per hectare. Costs of herbicide application in Seedbank were variable because of di ff erences in seedbank responses at each site. The range of costs for Seedbank was $ 55.74 – $ 289.28 Table 1. Percent cover of plant groups detected (mean ± SE) in 3 early successional plant community treatments across all study sites (n = 15) in Tennessee and Alabama, USA, June – August 2018. Treatment Life form Control Seedbank Planted F 2,28 P Bramble a 9 ± 2 A 10 ± 2 A 12 ± 2 A 2.35 0.114 Forb 48 ± 3 B 72 ± 2 A 64 ± 3 A 7.53 0.002 Grass 92 ± 2 A 63 ± 3 B 76 ± 3 B 10.96 ≤ 0.001 NWSG b 33 ± 4 C 49 ± 4 B 61 ± 3 A 15.11 ≤ 0.001 Woody c 9 ± 2 A 7 ± 1 A 7 ± 1 A 0.57 0.575 a Row means with the same letter were not di ff erent ( α = 0.05). b NWSG = native warm ‐ season grass. c Woody = shrubs, trees, and woody vines. Harper et al. • Establishing Early Successional Plant Communities 623 23285540, 2021, 4, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1232 by University Of Florida, Wiley Online Library on [06/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License per hectare and averaged $ 126.69 per hectare, including the initial $ 20.26 per hectare glyphosate application. On average, Planted units required 0.4 (SE = 0.11; range 0 – 3) entries per site per year (excluding the initial herbicide treatment to control tall fescue) for herbicide applications, whereas Seedbank units required 1.3 (SE = 0.18; range 0 – 4) entries per site per year. DISCUSSION Our study documents similar results following 2 approaches to restore a native plant community on sites previously dominated by a nonnative grass and indicated that planting native grasses and forbs is not necessary on a majority of sites when the objective is to improve those sites for grass - land songbirds, bobwhite, wild turkey, and white ‐ tailed deer. We measured treatment e ff ects in fi elds approximately 2 – 5 ha in size, but we stress that managers can use these results at the appropriate management scale while consid - ering the area requirements of focal species to simulta - neously bene fi t multiple conservation ‐ priority species as well as the 2 most ‐ popular game species in the eastern U.S. In particular, eradicating tall fescue with a single herbicide application and allowing the seedbank to respond without planting anything more than doubled nutrition available for white ‐ tailed deer, improved potential brooding cover for turkeys and quail, and provided structure consistent with that selected by several grassland birds. Plant composition and structure in Seedbank and Planted treatments were similar despite the use of di ff erent estab - lishment approaches. However, openness at ground level was greatest in Seedbank, and native grasses were more prevalent in Planted, which supported our hypotheses re - lated to structure. Forage availability for deer did not di ff er between Planted and Seedbank, which did not support our hypothesis, but the majority of deer forage in Planted was from species germinating from the seedbank. Plant phenology is an important consideration when making any herbicide application, and November glyph - osate applications e ff ectively controlled tall fescue. Long ‐ term commitments often required to control nonnative in - vasive species can be a discouraging factor for wildlife managers. However, similar to Harper and Gruchy (2009), our data indicated that tall fescue can be controlled with a single herbicide application made after the initial frosts in autumn when most desirable warm ‐ season plants are dead or dormant. Smith (1989) also reported better control of tall fescue with fall applications than spring applications. Sub - sequent spot ‐ spray applications greatly a ff ect restoration success by controlling undesirable competition as it occurs. Whether planting or using the seedbank to revegetate without planting, controlling competing vegetation is requisite to restoration success (Mitchell and Britton 2000, Bakker et al. 2003, Harper et al. 2007). One advantage of using the seedbank for revegetation is that a wide variety of herbicides are available to control undesirable species, Table 2. Vegetation pro fi le board estimates (mean ± SE) by treatment for individual strata at all study sites (n = 15) in Tennessee and Alabama, USA, June – August 2018. Treatment 0 – 25 cm a 25 – 50 cm 50 – 100 cm 100 – 150 cm 150 – 200 cm Control 100 ± 0.2 A 88 ± 2 B 60 ± 3 B 35 ± 3 B 20 ± 3 B Seedbank 99 ± 0.6 A 95 ± 1 A 77 ± 3 A 54 ± 3 A 34 ± 3 AB Planted 100 ± 0.1 A 99 ± 1 A 82 ± 2 A 53 ± 3 A 37 ± 3 A F 2,28 1.70 11.17 6.82 4.52 4.45 P 0.201 0.004 0.004 0.020 0.021 a Column means with the same letter are not di ff erent ( α = 0.05). Table 3. Plant species determined to be moderate ‐ to highly ‐ selected deer forages by selectivity index (Chesson 1983) across all study sites (n = 15) in Tennessee and Alabama, USA, June – August 2017 – 2018. Common name Scienti fi c name Life form IV a CP% b Common hackberry Celtis occidentalis Tree 0.034 11.9 Sti ff ticktrefoil Desmodium obtusum Forb 0.025 19.7 Common selfheal Prunella vulgaris Forb 0.025 12.1 Old ‐ fi eld aster Symphyotrichum pilosum Forb 0.022 14.7 American pokeweed Phytolacca americana Forb 0.014 28.0 Trumpet creeper Campsis radicans Vine 0.013 12.6 Panicled ‐ leaf ticktrefoil Desmodium paniculatum Forb 0.011 17.0 Ticktrefoil Desmodium spp Forb 0.011 18.4 Aster Symphyotrichum spp Forb 0.009 14.7 Northern dewberry Rubus fl agellaris Bramble 0.008 10.6 Red clover Trifolium pretense Forb 0.008 21.6 Common persimmon Diospyros virginiana Tr